Development and pseudovalidation of a questionnaire of frequency of drinks alcoholics for a Colombian population

To develop a questionnaire of frequency of consumptionof alcoholic beverages (FFQ) useful to classify subject according tointake of alcohol. Place and date. Bucaramanga, Colombia. June2002-September 2004. Methods. Hundred nine adults (20 to 60years) selected randomly completed a daily record during 12 serialweeks of their intake of alcohol (R12). The alcoholic drinks for theFFQ were selected by multiple lineal regression and Max_r. Thealcoholic drinks for the check list were selected by multiple linealregression and Max_r. FFQ′s for each subject was completed bythe authors with base in the consumption frequency reported in theR12. The correlation and the agreement between the FFQ and theR12 were valued using correlation coeffi cients of Spearman (rs) andthe half difference and you limit them of agreement of Bland andAltman (Bias-LOA). Results. Five drinks alcoholics were selectedfor the check list of the FFQ. Seven FFQ, with different combinationsof the fi ve selected drinks was evaluated. The FFQ′s and R12 werecorrelated well. The rs was between 0,58 and 0,68. The half inferiorlimit in the bias was of 3,3% and the superior of 91,9%. However,there was no indication that the degree of agreement varied with thelevel of intake. Conclusions. The FFQ with the fi ve drinks selectedoriginally can be useful to classify subject of agreement with its level of consumption of alcohol. Their performance in unique evaluation for individual and in different population’s groups it should be considered carefully. We justify the later realization of a classic study of reliability and validity.